USB4 2.0 Cables Capable of 80 Gbps Data and Power Delivery of 60 W and 240 W, Get Certified

[TheLostSwede]

Here are a couple I looked at, but regardless, 3800MB/s is approximately 30Gbps.....10Gbps overhead seems a bit steep....I don't know, just haven't seen anything convincing that it's using the 40Gbps for data

Well, we have one third party host controller and one bridge chip so far. It's possible that ASMedia didn't get it all right at its first attempt. I don't know if AMD's implementation in their mobile chips is better or not. We also have Qualcomm now, but not seen anyone test the USB4 performance.
That said, Thunderbolt 3/4 are the same, at around 9 Gbps behind the claimed 32 Gbps data rate.
It'll be interesting to see what Thunderbolt 5 delivers in terms of performance.

Others got the Adata drive to go faster, even if it's slow on the write side.

ADATA SE920 2TB USB4 Portable SSD Review - Fastest ever external drive tested

ADATA's SE920 USB4 SSD is the fastest and most powerful bus-powered pocket portable external storage device we've ever tested, capable of up to 3,800 MB/s.

www.tweaktown.com

 

[Wirko]

For PCIe devices such as eGPUs and M.2 SSDs in enclosures, I'd expect PCIe tunneling to be used for simplicity and cost saving. You need no electronics for that to work, or just redrivers in the worst case. But the speed is limited to 32 Gbit/s of course.

 

[Nhonho]

More nonsense nomenclature

This is the only system we need:

View attachment 365923
etc...

This is the USB Gen 1 (USB 1.x) trident logo:

The USB 2.0 trident logo is the same as the first gen with a "+":

The biggest missed opportunity in the past 8 years of USB is not being able to link and daisy chain fans and RGB devices together with USB.

It would be great if all the fans and LEDs were connected via internal USB-C connectors. I hate LEDs on PCs (I turn them all off) and this would make it much easier to turn on, off, change the colors of all the LEDs on the PC and also change the fan speed, via BIOS or via app.

USB4 2.0... I thought they were using the 'Gen2x#' moniker?
Guess I'll just pedantically memorize the latest USB-IF nomenclature

I only pay attention to the speed of the USB port, how many Watts it provides (and whether it supports PD) and whether it transmits DisplayPort signal. I don't pay attention to those confusing and useless names from USB-IF.

With PC motherboard manufacturers implementing 40 Gbps USB4, one wonders what's next for the [mostly] universal connectivity standard. Turns out, it is USB4 2.0, which yields the connector 80 Gbps of data bandwidth per direction, and 60 W of power-delivery (PD). Power delivery for USB4 2.0 can be as high as 240 W (48 V, 5 A). Japanese PC peripherals maker Elecom is the first with certified cables for both 80 Gbps + 60 W PD and 80 Gbps + 240 W PD. Besides power and data, both cables support DisplayPort passthrough for up to 8K @ 60 Hz (7680 x 4320 pixels). USB-IF, the governing body of the USB standard, is expected to formally launch USB4 2.0 in December 2024. A typical USB4 2.0 host controller will require at least a PCI-Express 5.0 x4 system bus connection to reach the standard's maximum bandwidth.

Does it still make sense to exist USB-C Thunderbolt ports, since the new USB-C ports already have faster data transmission speeds than USB-C Thunderbolt ports and also support PD and DisplayPort?? Regular USB-C ports (non-Thunderbolt) with 120gbps data transmission speeds will be released soon.

The USB-C connectors are too expensive to do that.

I think it would only cost a few cents more.

 

[Darmok N Jalad]

We also have Qualcomm now, but not seen anyone test the USB4 performance.

According to Paul, Qualcomm was quick to say 40Gbps is possible on any SD X hardware. He didn't test the claim that I could find.

 

[LabRat 891]

The biggest missed opportunity in the past 8 years of USB is not being able to link and daisy chain fans and RGB devices together with USB.

'Love the concept, but it's just not something I could see ever becoming anything more than an informal standard. Especially w/ USB never focusing on daisy chaining.
TBQH, that idea could've been (arguably, should've been) the last gasp long-term survival plan for IEEE 1394 FireWire in the consumer market.

The power levels, voltage levels, and inherit daisy chaining, all lend themselves well.

Alternatively, there's CANBUS, Single Pair PoE 10/100, and various Automotive Standard(s) Ethernet...

 

[Wirko]

The USB-C connectors are too expensive to do that.
All the other connectors are only allowed on one end of the cable to be allowed to be called USB.

Cables with micro-A at one end and micro-B at the other are standard. (I think of micro USB 2.0, not those laughable wide 3.0.)
But the problem with A, B and C is also that the connectors are quite chunky, with strain relief made for everyday handling. Cables are also more rigid than individual wires. You can bend wires on 3- or 4-pin fan cables right next to the connector to your will, in any direction.

 

[Easo]

I had to do a double take on seeing that 2.0.

They really screwed up with the naming.

 

[TheLostSwede]

For PCIe devices such as eGPUs and M.2 SSDs in enclosures, I'd expect PCIe tunneling to be used for simplicity and cost saving. You need no electronics for that to work, or just redrivers in the worst case. But the speed is limited to 32 Gbit/s of course.

40 Gbps with USB4, still.

I think it would only cost a few cents more.

USB-C connectors are actually quite complex an expensive if that was what the plan to be used in this scenario. They all have a PCB as part of cable connectors and the tolerances are quite strict.
The cheap pin-header typ connectors are by far the cheapest connectors out there, which is why they're so widely used.

According to Paul, Qualcomm was quick to say 40Gbps is possible on any SD X hardware. He didn't test the claim that I could find.

Yeah, I was trying to find some tests as well, but couldn't find any.

Cables with micro-A at one end and micro-B at the other are standard. (I think of micro USB 2.0, not those laughable wide 3.0.)

Show me such a cable? I'm aware that it was part of the standard at one point, but I never saw such a cable in my entire life. A couple of devices had the micro-B connector though.

But the problem with A, B and C is also that the connectors are quite chunky, with strain relief made for everyday handling. Cables are also more rigid than individual wires. You can bend wires on 3- or 4-pin fan cables right next to the connector to your will, in any direction.

Indeed.

 

[Wirko]

40 Gbps with USB4, still.

Yeah, that was a bit hard to understand to me. "Tunneling" means wrapping PCIe packets in USB4 packets. It's not a PCIe exclusive mode, which would simply send PCIe data over the wires and be limited to PCIe standard speeds. (It would also enable operation without a bridge chip in the peripheral device).

Is there significant USB4 protocol overhead in tunneling? One source at Reddit says that "ASM2464 is fully bottlenecked by the 40G TB/USB4 connection: NVMes often reach ~3800 MiB/s. This is pretty much the limit of the current 128 Byte payload PCIe through USB4 40G connections."

https://www.reddit.com/r/framework/comments/1co0xcc

 

[Nhonho]

USB-C connectors are actually quite complex an expensive if that was what the plan to be used in this scenario.

I don't think so.

They all have a PCB as part of cable connectors

Only active cables are this way.

 

[Ravenmaster]

Blazin USB 4 2.0

You just know the guy who named that thing was a pot head

 

[Prima.Vera]

God damn USB forum with their utterly retarded and over complicated naming schemes! 'da fruck is this USB4 2.0 now? Why not USB4.2 or something? Or USB4-80G, much simpler and better to understand.
Damn, those guys naming this protocol must have work for Intel. Overcomplicating and retarded naming conventions are their thing.

 

[TheLostSwede]

Yeah, that was a bit hard to understand to me. "Tunneling" means wrapping PCIe packets in USB4 packets. It's not a PCIe exclusive mode, which would simply send PCIe data over the wires and be limited to PCIe standard speeds. (It would also enable operation without a bridge chip in the peripheral device).

Is there significant USB4 protocol overhead in tunneling? One source at Reddit says that "ASM2464 is fully bottlenecked by the 40G TB/USB4 connection: NVMes often reach ~3800 MiB/s. This is pretty much the limit of the current 128 Byte payload PCIe through USB4 40G connections."

https://www.reddit.com/r/framework/comments/1co0xcc

Only USB4 2.0 as it's so elegantly named, changes the encoding to PAM-3, so USB4 still has the same overheads as USB 3.2 with 128/132-bit encoding. I guess this is in part why the performance figures aren't as good as they should be.

I don't think so.
Only active cables are this way.

Well, feel free to think so, but all cables have a PCB that the wires are soldered onto, even USB-A/B/micro is like that. If they don't, they won't last 5 minutes.
However, the complexity of USB-C is a lot higher due to there being so many more cables involved, makes it a lot more expensive, unless it's a USB 2.0 only cable with 3A power limit.

I mean, you have 22/24-wires in a full USB-C connector vs. four or nine for USB-A (2.0 vs 3.0). That's a lot of extra complexity, which equals extra cost, not even taking the extra wires into consideration.

God damn USB forum with their utterly retarded and over complicated naming schemes! 'da fruck is this USB4 2.0 now? Why not USB4.2 or something? Or USB4-80G, much simpler and better to understand.
Damn, those guys naming this protocol must have work for Intel. Overcomplicating and retarded naming conventions are their thing.

The marketing name is USB 80Gbps.

And yes, Intel is one of the companies on the board of the USB-IF that is deciding on the various specs.

 

[Prima.Vera]

Wait, Abdul is both Chairman and CTO?? And name pusher?

 

[TumbleGeorge]

Wait, Abdul is both Chairman and CTO?? And name pusher?

Must re-read the lists. There has more repeats.

 

[Nhonho]

Well, feel free to think so, but all cables have a PCB that the wires are soldered onto, even USB-A/B/micro is like that. If they don't, they won't last 5 minutes.
However, the complexity of USB-C is a lot higher due to there being so many more cables involved, makes it a lot more expensive, unless it's a USB 2.0 only cable with 3A power limit.

I mean, you have 22/24-wires in a full USB-C connector vs. four or nine for USB-A (2.0 vs 3.0). That's a lot of extra complexity, which equals extra cost, not even taking the extra wires into consideration.

You have the freedom to write whatever wrong things you want...

 

[TheLostSwede]

You have the freedom to write whatever wrong things you want...

Ok, please provide a mechanical drawing/picture of a USB-C connector, for cables, that doesn't have a PCB inside them then.
Not just the header, but the full connector with wire interface.

I think you'll find that they all look something like this.

Or this

Neither kind above, is for 5A power and thus doesn't incorporate the E-Marker.

 

[TumbleGeorge]

I don't know why you are complaining, for something that can be bought for $78-80 for a box of 100 inside(~$0.8, per piece) on amazon and cheaper from Chinese stores, including the circuit boards and plug sheets (connector). And these prices are nowhere near wholesale prices for millions of units and production costs. It was expensive, yes, for the special directional cables, for audiophiles and moviegoers.

 

[TheLostSwede]

I don't know why you are complaining, for something that can be bought for $78-80 for a box of 100 inside(~$0.8, per piece) on amazon and cheaper from Chinese stores, including the circuit boards and plug sheets (connector). And these prices are nowhere near wholesale prices for millions of units and production costs. It was expensive, yes, for the special directional cables, for audiophiles and moviegoers.

Complaining? I'm saying they're not going to be used for fans or RGB due to the high cost vs. pin-headers, but apparently I'm wrong, so...
Even if they're 50 cents, a plastic pin header is 2 cents, that's 25x the cost.

The ones for 80 cents on Amazon also appear to be limited to USB 2.0 and can't do C-C according to the comments, although I guess that might be someone that didn't understand that the cables are directional unless you solder a resistor or have chip that detects how the cable has been plugged in. Also, they apparently do PD, but I guess that's not relevant to this conversation.

 

[TumbleGeorge]

Must be Google translate guilty. Sorry! But in any case no matter than 25x difference price still more than enough cheap.

 

[TheLostSwede]

Must be Google translate guilty. Sorry! But in any case no matter than 25x difference price still more than enough cheap.

Not when you make a million of something that you want to either sell at the lowest possible price, or make the largest possible margin out of.

 

[Bagerklestyne]

Question (I've dug and can't find anything)

Any ideas what the maximum high refresh rates (above 60hz) alt modes can drive screens at ?

 

[Minus Infinity]

More nonsense nomenclature

This is the only system we need:

View attachment 365923
etc...

If they do PD, then that should be clearly stated on the ports and cables in Watts, not hidden behind some bullshit lookup tables of USB version, sub-version, and bogus caveats* or using the words "up to" everywhere. I can type at "up to" 300 words a minute and my monitor's pixel response is 1ms*

View attachment 365925

Why is that so hard?

IMO it should be:

USB 3.0 5Gb/s
USB 3.1 10Gb/s
USB 3.2 20 Gb/s

That would make it perfectly clear what you are getting

BTW is it possible for USB PD cables to offer better than USB 2 data speed?

 

[R-T-B]

BTW is it possible for USB PD cables to offer better than USB 2 data speed?

Certainly yes as products have been shipped fitting that description?

 

[Chrispy_]

Certainly yes as products have been shipped fitting that description?

The problem is that the cable industry is a minefield. So many "USB PD charging cables" are just intended for charging only and offer the bare minimum data connectivity.

If you want to buy a cable that's both charging and high speed data transfer, you really need to hunt through a lot of other cables wherever you're shopping because every manufacturer makes and sells 20 different charging-only cables for each PD-capable high-bandwidth cable.

The masses charging their phones and tablets from a wall adapter vastly outnumber those who are trying to use aftermarket USB cables to transfer data from fast devices. WiFi and ethernet are the default data transfer mechanisms for most people and half the population doesn't even realise USB can be used for data transfer at all!